init

added_tokens.json

@@ -0,0 +1 @@
+{"<sep/>": 57522, "<s_iitcdip>": 57523, "<s_synthdog>": 57524, "▁□": 57525, "▁△": 57526, "▁●": 57527, "▁〆": 57528, "▁〇": 57529, "▁〱": 57530, "▁〲": 57531, "▁〻": 57532, "▁〽": 57533, "▁ぜ": 57534, "▁ぞ": 57535, "▁づ": 57536, "▁ぱ": 57537, "▁ぴ": 57538, "▁ぷ": 57539, "▁ゃ": 57540, "▁ゅ": 57541, "▁ょ": 57542, "▁ゑ": 57543, "▁ゝ": 57544, "▁ゟ": 57545, "▁ィ": 57546, "▁ゼ": 57547, "▁ヅ": 57548, "▁ヌ": 57549, "▁ユ": 57550, "▁ョ": 57551, "▁ヨ": 57552, "▁ヱ": 57553, "▁ヲ": 57554, "▁ヽ": 57555, "▁㊀": 57556, "▁㊁": 57557, "▁㊂": 57558, "▁㊃": 57559, "▁㊄": 57560, "▁㐂": 57561, "▁㐫": 57562, "▁㑹": 57563, "▁㒵": 57564, "▁㓛": 57565, "▁㔟": 57566, "▁㔫": 57567, "▁㔺": 57568, "▁㕝": 57569, "▁㖧": 57570, "▁㘴": 57571, "▁㚑": 57572, "▁㝡": 57573, "▁㠶": 57574, "▁㡌": 57575, "▁㡣": 57576, "▁㣲": 57577, "▁㤙": 57578, "▁㫖": 57579, "▁㮇": 57580, "▁㱕": 57581, "▁䂓": 57582, "▁䏣": 57583, "▁䏻": 57584, "▁䐈": 57585, "▁䑓": 57586, "▁䕃": 57587, "▁䕲": 57588, "▁䱡": 57589, "▁䲙": 57590, "▁丈": 57591, "▁丞": 57592, "▁丱": 57593, "▁乂": 57594, "▁亍": 57595, "▁仇": 57596, "▁仏": 57597, "▁仗": 57598, "▁仭": 57599, "▁仮": 57600, "▁仰": 57601, "▁伏": 57602, "▁伐": 57603, "▁伱": 57604, "▁伽": 57605, "▁佀": 57606, "▁佞": 57607, "▁侘": 57608, "▁侭": 57609, "▁侮": 57610, "▁侲": 57611, "▁俎": 57612, "▁俤": 57613, "▁俯": 57614, "▁俵": 57615, "▁倦": 57616, "▁倭": 57617, "▁倰": 57618, "▁偖": 57619, "▁偶": 57620, "▁偸": 57621, "▁傅": 57622, "▁傘": 57623, "▁傲": 57624, "▁傾": 57625, "▁僂": 57626, "▁僉": 57627, "▁僑": 57628, "▁僜": 57629, "▁兇": 57630, "▁兎": 57631, "▁兜": 57632, "▁兪": 57633, "▁兮": 57634, "▁冥": 57635, "▁冶": 57636, "▁凄": 57637, "▁凍": 57638, "▁処": 57639, "▁凧": 57640, "▁凩": 57641, "▁凶": 57642, "▁刃": 57643, "▁刅": 57644, "▁刕": 57645, "▁刪": 57646, "▁刹": 57647, "▁剄": 57648, "▁削": 57649, "▁劄": 57650, "▁劣": 57651, "▁努": 57652, "▁劫": 57653, "▁励": 57654, "▁労": 57655, "▁勃": 57656, "▁勦": 57657, "▁勿": 57658, "▁匁": 57659, "▁匕": 57660, "▁匙": 57661, "▁匜": 57662, "▁匠": 57663, "▁匹": 57664, "▁卑": 57665, "▁卒": 57666, "▁卓": 57667, "▁卥": 57668, "▁卩": 57669, "▁卸": 57670, "▁厄": 57671, "▁厠": 57672, "▁厭": 57673, "▁叉": 57674, "▁叙": 57675, "▁叛": 57676, "▁叩": 57677, "▁吁": 57678, "▁吏": 57679, "▁吟": 57680, "▁吾": 57681, "▁呈": 57682, "▁呉": 57683, "▁呵": 57684, "▁咲": 57685, "▁咳": 57686, "▁咸": 57687, "▁咽": 57688, "▁哀": 57689, "▁哢": 57690, "▁哲": 57691, "▁唄": 57692, "▁唯": 57693, "▁啁": 57694, "▁啌": 57695, "▁啓": 57696, "▁啖": 57697, "▁啗": 57698, "▁啜": 57699, "▁喚": 57700, "▁喞": 57701, "▁喟": 57702, "▁喪": 57703, "▁営": 57704, "▁嗔": 57705, "▁嗜": 57706, "▁嘗": 57707, "▁嘘": 57708, "▁嘲": 57709, "▁噀": 57710, "▁噫": 57711, "▁噲": 57712, "▁嚼": 57713, "▁囀": 57714, "▁囁": 57715, "▁囓": 57716, "▁囚": 57717, "▁囲": 57718, "▁圉": 57719, "▁圑": 57720, "▁坂": 57721, "▁垂": 57722, "▁垣": 57723, "▁埵": 57724, "▁堀": 57725, "▁堤": 57726, "▁堰": 57727, "▁堵": 57728, "▁塀": 57729, "▁塒": 57730, "▁塚": 57731, "▁塩": 57732, "▁塵": 57733, "▁壇": 57734, "▁壊": 57735, "▁壒": 57736, "▁壜": 57737, "▁壼": 57738, "▁夀": 57739, "▁央": 57740, "▁夷": 57741, "▁奚": 57742, "▁奠": 57743, "▁奢": 57744, "▁妃": 57745, "▁妄": 57746, "▁妨": 57747, "▁姚": 57748, "▁姜": 57749, "▁姨": 57750, "▁姪": 57751, "▁姫": 57752, "▁娌": 57753, "▁娟": 57754, "▁娥": 57755, "▁娵": 57756, "▁娶": 57757, "▁婉": 57758, "▁婕": 57759, "▁媚": 57760, "▁媱": 57761, "▁媵": 57762, "▁媽": 57763, "▁嫉": 57764, "▁嫗": 57765, "▁嫡": 57766, "▁嬖": 57767, "▁嬢": 57768, "▁嬶": 57769, "▁孁": 57770, "▁孩": 57771, "▁宛": 57772, "▁宦": 57773, "▁寓": 57774, "▁寔": 57775, "▁寤": 57776, "▁寵": 57777, "▁寸": 57778, "▁専": 57779, "▁尉": 57780, "▁尠": 57781, "▁尢": 57782, "▁尨": 57783, "▁尭": 57784, "▁屁": 57785, "▁屍": 57786, "▁屐": 57787, "▁屠": 57788, "▁屡": 57789, "▁履": 57790, "▁屩": 57791, "▁屯": 57792, "▁岐": 57793, "▁岑": 57794, "▁岨": 57795, "▁峠": 57796, "▁峽": 57797, "▁崔": 57798, "▁崖": 57799, "▁嶺": 57800, "▁嶼": 57801, "▁巌": 57802, "▁巵": 57803, "▁巻": 57804, "▁帖": 57805, "▁帙": 57806, "▁帥": 57807, "▁帽": 57808, "▁庁": 57809, "▁庇": 57810, "▁庻": 57811, "▁廃": 57812, "▁廊": 57813, "▁廓": 57814, "▁廩": 57815, "▁廪": 57816, "▁廰": 57817, "▁廱": 57818, "▁廷": 57819, "▁廻": 57820, "▁廼": 57821, "▁廾": 57822, "▁弉": 57823, "▁弑": 57824, "▁弓": 57825, "▁弖": 57826, "▁弘": 57827, "▁弦": 57828, "▁彂": 57829, "▁彫": 57830, "▁彰": 57831, "▁彳": 57832, "▁徂": 57833, "▁径": 57834, "▁従": 57835, "▁徤": 57836, "▁徳": 57837, "▁徹": 57838, "▁徽": 57839, "▁忝": 57840, "▁忰": 57841, "▁怖": 57842, "▁怜": 57843, "▁怡": 57844, "▁怨": 57845, "▁恍": 57846, "▁恕": 57847, "▁恖": 57848, "▁恚": 57849, "▁恟": 57850, "▁恥": 57851, "▁恰": 57852, "▁悉": 57853, "▁悋": 57854, "▁悒": 57855, "▁悔": 57856, "▁悦": 57857, "▁悧": 57858, "▁悩": 57859, "▁悶": 57860, "▁惜": 57861, "▁惟": 57862, "▁惹": 57863, "▁惻": 57864, "▁愀": 57865, "▁愁": 57866, "▁愍": 57867, "▁愚": 57868, "▁愧": 57869, "▁愽": 57870, "▁慂": 57871, "▁慄": 57872, "▁慎": 57873, "▁慕": 57874, "▁慙": 57875, "▁慝": 57876, "▁慣": 57877, "▁慥": 57878, "▁慨": 57879, "▁慳": 57880, "▁慾": 57881, "▁憖": 57882, "▁憤": 57883, "▁憲": 57884, "▁懈": 57885, "▁懐": 57886, "▁懶": 57887, "▁戌": 57888, "▁戟": 57889, "▁扇": 57890, "▁扑": 57891, "▁扨": 57892, "▁扱": 57893, "▁抄": 57894, "▁抔": 57895, "▁択": 57896, "▁抦": 57897, "▁披": 57898, "▁担": 57899, "▁拊": 57900, "▁拌": 57901, "▁拓": 57902, "▁拝": 57903, "▁拵": 57904, "▁挊": 57905, "▁挌": 57906, "▁挙": 57907, "▁挫": 57908, "▁挽": 57909, "▁捉": 57910, "▁捏": 57911, "▁捜": 57912, "▁捧": 57913, "▁捨": 57914, "▁捫": 57915, "▁捲": 57916, "▁掃": 57917, "▁掘": 57918, "▁掠": 57919, "▁掩": 57920, "▁掫": 57921, "▁掴": 57922, "▁掾": 57923, "▁揮": 57924, "▁搏": 57925, "▁搦": 57926, "▁摘": 57927, "▁摚": 57928, "▁摠": 57929, "▁摧": 57930, "▁撃": 57931, "▁撈": 57932, "▁撮": 57933, "▁撲": 57934, "▁撹": 57935, "▁擬": 57936, "▁擲": 57937, "▁擴": 57938, "▁攀": 57939, "▁敦": 57940, "▁斎": 57941, "▁斐": 57942, "▁斤": 57943, "▁旃": 57944, "▁旌": 57945, "▁旦": 57946, "▁旨": 57947, "▁旬": 57948, "▁旭": 57949, "▁昇": 57950, "▁昏": 57951, "▁昨": 57952, "▁昭": 57953, "▁昴": 57954, "▁晩": 57955, "▁暫": 57956, "▁暮": 57957, "▁曇": 57958, "▁曚": 57959, "▁曜": 57960, "▁曝": 57961, "▁曰": 57962, "▁曹": 57963, "▁朋": 57964, "▁朴": 57965, "▁朶": 57966, "▁杉": 57967, "▁杏": 57968, "▁杓": 57969, "▁杙": 57970, "▁杢": 57971, "▁杣": 57972, "▁杲": 57973, "▁杼": 57974, "▁枕": 57975, "▁枯": 57976, "▁枲": 57977, "▁柄": 57978, "▁柑": 57979, "▁柤": 57980, "▁柩": 57981, "▁栗": 57982, "▁栝": 57983, "▁栫": 57984, "▁栲": 57985, "▁栽": 57986, "▁桓": 57987, "▁桝": 57988, "▁桴": 57989, "▁桾": 57990, "▁桿": 57991, "▁梓": 57992, "▁梺": 57993, "▁梼": 57994, "▁棊": 57995, "▁棍": 57996, "▁棐": 57997, "▁棚": 57998, "▁棟": 57999, "▁棠": 58000, "▁棹": 58001, "▁椅": 58002, "▁椽": 58003, "▁楑": 58004, "▁楙": 58005, "▁楜": 58006, "▁楝": 58007, "▁楡": 58008, "▁楪": 58009, "▁楫": 58010, "▁楴": 58011, "▁榊": 58012, "▁榎": 58013, "▁榑": 58014, "▁榛": 58015, "▁榧": 58016, "▁榪": 58017, "▁榾": 58018, "▁榿": 58019, "▁槁": 58020, "▁槍": 58021, "▁槖": 58022, "▁槫": 58023, "▁槹": 58024, "▁樒": 58025, "▁樔": 58026, "▁樽": 58027, "▁橘": 58028, "▁橡": 58029, "▁橦": 58030, "▁檋": 58031, "▁檍": 58032, "▁檗": 58033, "▁櫂": 58034, "▁櫨": 58035, "▁櫺": 58036, "▁欝": 58037, "▁欬": 58038, "▁歟": 58039, "▁歴": 58040, "▁殊": 58041, "▁殖": 58042, "▁殞": 58043, "▁殷": 58044, "▁殽": 58045, "▁毆": 58046, "▁毎": 58047, "▁毟": 58048, "▁毫": 58049, "▁毯": 58050, "▁汚": 58051, "▁汲": 58052, "▁沢": 58053, "▁沸": 58054, "▁沼": 58055, "▁沽": 58056, "▁泄": 58057, "▁泅": 58058, "▁泔": 58059, "▁泝": 58060, "▁泣": 58061, "▁泳": 58062, "▁洒": 58063, "▁浅": 58064, "▁浜": 58065, "▁浥": 58066, "▁淋": 58067, "▁淚": 58068, "▁淪": 58069, "▁淫": 58070, "▁済": 58071, "▁渓": 58072, "▁渟": 58073, "▁渣": 58074, "▁渾": 58075, "▁溙": 58076, "▁溝": 58077, "▁滓": 58078, "▁漁": 58079, "▁漉": 58080, "▁漬": 58081, "▁漻": 58082, "▁漿": 58083, "▁潅": 58084, "▁潯": 58085, "▁濤": 58086, "▁濯": 58087, "▁瀬": 58088, "▁瀰": 58089, "▁灘": 58090, "▁烮": 58091, "▁烹": 58092, "▁焔": 58093, "▁焼": 58094, "▁煆": 58095, "▁煉": 58096, "▁煩": 58097, "▁煬": 58098, "▁煻": 58099, "▁煽": 58100, "▁熇": 58101, "▁熬": 58102, "▁燗": 58103, "▁燠": 58104, "▁燵": 58105, "▁燼": 58106, "▁爈": 58107, "▁爨": 58108, "▁爪": 58109, "▁爹": 58110, "▁牝": 58111, "▁牡": 58112, "▁犇": 58113, "▁狄": 58114, "▁狎": 58115, "▁狐": 58116, "▁狙": 58117, "▁狩": 58118, "▁狭": 58119, "▁狷": 58120, "▁狻": 58121, "▁猨": 58122, "▁猩": 58123, "▁猴": 58124, "▁猶": 58125, "▁獅": 58126, "▁獏": 58127, "▁獬": 58128, "▁玄": 58129, "▁珊": 58130, "▁珪": 58131, "▁琢": 58132, "▁琫": 58133, "▁瓊": 58134, "▁瓠": 58135, "▁甀": 58136, "▁甍": 58137, "▁甑": 58138, "▁甫": 58139, "▁畋": 58140, "▁畏": 58141, "▁畑": 58142, "▁畔": 58143, "▁畷": 58144, "▁畸": 58145, "▁畾": 58146, "▁疇": 58147, "▁疋": 58148, "▁疳": 58149, "▁疾": 58150, "▁痘": 58151, "▁痴": 58152, "▁瘖": 58153, "▁瘞": 58154, "▁瘡": 58155, "▁瘰": 58156, "▁癈": 58157, "▁癪": 58158, "▁癸": 58159, "▁皂": 58160, "▁皹": 58161, "▁皿": 58162, "▁盧": 58163, "▁盲": 58164, "▁眇": 58165, "▁眛": 58166, "▁眤": 58167, "▁眦": 58168, "▁眺": 58169, "▁睦": 58170, "▁矚": 58171, "▁矢": 58172, "▁矩": 58173, "▁砂": 58174, "▁碁": 58175, "▁碓": 58176, "▁碧": 58177, "▁磤": 58178, "▁磽": 58179, "▁礒": 58180, "▁礬": 58181, "▁祐": 58182, "▁禅": 58183, "▁禊": 58184, "▁禦": 58185, "▁秡": 58186, "▁秬": 58187, "▁稗": 58188, "▁稚": 58189, "▁窩": 58190, "▁窮": 58191, "▁窶": 58192, "▁窹": 58193, "▁竃": 58194, "▁竜": 58195, "▁竪": 58196, "▁竭": 58197, "▁笄": 58198, "▁笊": 58199, "▁笘": 58200, "▁笞": 58201, "▁笧": 58202, "▁筌": 58203, "▁筎": 58204, "▁筑": 58205, "▁筥": 58206, "▁筧": 58207, "▁筬": 58208, "▁筮": 58209, "▁箆": 58210, "▁箒": 58211, "▁箙": 58212, "▁箭": 58213, "▁箴": 58214, "▁範": 58215, "▁築": 58216, "▁篗": 58217, "▁篼": 58218, "▁簀": 58219, "▁簑": 58220, "▁簒": 58221, "▁簓": 58222, "▁簔": 58223, "▁簗": 58224, "▁簟": 58225, "▁簪": 58226, "▁簾": 58227, "▁籃": 58228, "▁籆": 58229, "▁籩": 58230, "▁籬": 58231, "▁粂": 58232, "▁粃": 58233, "▁粔": 58234, "▁粕": 58235, "▁粘": 58236, "▁粲": 58237, "▁粳": 58238, "▁糄": 58239, "▁糒": 58240, "▁糕": 58241, "▁糝": 58242, "▁糧": 58243, "▁糫": 58244, "▁糶": 58245, "▁糺": 58246, "▁紐": 58247, "▁紗": 58248, "▁紡": 58249, "▁紵": 58250, "▁紺": 58251, "▁絅": 58252, "▁絆": 58253, "▁緇": 58254, "▁緑": 58255, "▁緒": 58256, "▁緡": 58257, "▁縁": 58258, "▁縊": 58259, "▁縑": 58260, "▁縛": 58261, "▁縡": 58262, "▁縢": 58263, "▁縧": 58264, "▁縺": 58265, "▁縻": 58266, "▁繅": 58267, "▁繆": 58268, "▁繊": 58269, "▁繕": 58270, "▁繞": 58271, "▁繻": 58272, "▁纔": 58273, "▁纜": 58274, "▁罟": 58275, "▁罵": 58276, "▁罷": 58277, "▁羙": 58278, "▁羞": 58279, "▁翌": 58280, "▁翠": 58281, "▁翫": 58282, "▁翰": 58283, "▁翳": 58284, "▁翺": 58285, "▁翼": 58286, "▁聢": 58287, "▁肇": 58288, "▁肚": 58289, "▁肢": 58290, "▁胝": 58291, "▁胤": 58292, "▁胴": 58293, "▁胼": 58294, "▁脔": 58295, "▁脩": 58296, "▁脾": 58297, "▁腎": 58298, "▁腨": 58299, "▁腫": 58300, "▁膂": 58301, "▁膏": 58302, "▁膝": 58303, "▁膳": 58304, "▁臂": 58305, "▁臆": 58306, "▁臍": 58307, "▁臑": 58308, "▁臙": 58309, "▁臚": 58310, "▁臣": 58311, "▁舁": 58312, "▁舂": 58313, "▁舅": 58314, "▁舎": 58315, "▁舟": 58316, "▁艘": 58317, "▁艚": 58318, "▁艡": 58319, "▁艪": 58320, "▁艫": 58321, "▁艶": 58322, "▁芋": 58323, "▁芍": 58324, "▁芙": 58325, "▁芟": 58326, "▁芡": 58327, "▁芦": 58328, "▁芭": 58329, "▁苗": 58330, "▁苛": 58331, "▁苡": 58332, "▁苧": 58333, "▁苪": 58334, "▁苵": 58335, "▁苺": 58336, "▁苻": 58337, "▁茄": 58338, "▁茘": 58339, "▁茨": 58340, "▁荅": 58341, "▁莕": 58342, "▁莚": 58343, "▁莟": 58344, "▁莨": 58345, "▁莾": 58346, "▁菊": 58347, "▁菔": 58348, "▁菖": 58349, "▁菝": 58350, "▁菩": 58351, "▁菪": 58352, "▁菫": 58353, "▁菰": 58354, "▁菱": 58355, "▁菽": 58356, "▁萄": 58357, "▁萆": 58358, "▁萍": 58359, "▁萓": 58360, "▁萩": 58361, "▁葍": 58362, "▁葎": 58363, "▁葛": 58364, "▁葜": 58365, "▁葭": 58366, "▁葶": 58367, "▁蒔": 58368, "▁蒜": 58369, "▁蒲": 58370, "▁蓍": 58371, "▁蓙": 58372, "▁蓮": 58373, "▁蓴": 58374, "▁蔀": 58375, "▁蔓": 58376, "▁蔕": 58377, "▁蔞": 58378, "▁蔦": 58379, "▁蕀": 58380, "▁蕃": 58381, "▁蕗": 58382, "▁蕢": 58383, "▁蕣": 58384, "▁蕷": 58385, "▁蕺": 58386, "▁蕾": 58387, "▁薁": 58388, "▁薊": 58389, "▁薑": 58390, "▁薙": 58391, "▁薜": 58392, "▁薮": 58393, "▁薷": 58394, "▁藐": 58395, "▁藷": 58396, "▁蘘": 58397, "▁蘡": 58398, "▁蘩": 58399, "▁蘰": 58400, "▁蘼": 58401, "▁蘿": 58402, "▁虐": 58403, "▁虬": 58404, "▁虯": 58405, "▁虹": 58406, "▁蚢": 58407, "▁蚩": 58408, "▁蚫": 58409, "▁蚰": 58410, "▁蚶": 58411, "▁蚸": 58412, "▁蛁": 58413, "▁蛄": 58414, "▁蛉": 58415, "▁蛕": 58416, "▁蛞": 58417, "▁蛥": 58418, "▁蛮": 58419, "▁蛯": 58420, "▁蛸": 58421, "▁蜂": 58422, "▁蜈": 58423, "▁蜋": 58424, "▁蜐": 58425, "▁蜑": 58426, "▁蜩": 58427, "▁蜾": 58428, "▁蝓": 58429, "▁蝕": 58430, "▁蝤": 58431, "▁蝦": 58432, "▁蝴": 58433, "▁蝶": 58434, "▁蝿": 58435, "▁螂": 58436, "▁螇": 58437, "▁螉": 58438, "▁螖": 58439, "▁螺": 58440, "▁蟇": 58441, "▁蟛": 58442, "▁蟟": 58443, "▁蟶": 58444, "▁蟷": 58445, "▁蟻": 58446, "▁蠃": 58447, "▁蠎": 58448, "▁蠕": 58449, "▁蠜": 58450, "▁蠢": 58451, "▁蠮": 58452, "▁衆": 58453, "▁衒": 58454, "▁衢": 58455, "▁衲": 58456, "▁衽": 58457, "▁衾": 58458, "▁袞": 58459, "▁袢": 58460, "▁袷": 58461, "▁袿": 58462, "▁裒": 58463, "▁裕": 58464, "▁裘": 58465, "▁裙": 58466, "▁裟": 58467, "▁裸": 58468, "▁裹": 58469, "▁褁": 58470, "▁褄": 58471, "▁褊": 58472, "▁褌": 58473, "▁褒": 58474, "▁襌": 58475, "▁襴": 58476, "▁襷": 58477, "▁覧": 58478, "▁覿": 58479, "▁觜": 58480, "▁討": 58481, "▁訓": 58482, "▁訟": 58483, "▁詈": 58484, "▁詠": 58485, "▁詮": 58486, "▁詰": 58487, "▁詵": 58488, "▁誂": 58489, "▁誅": 58490, "▁誑": 58491, "▁誓": 58492, "▁誕": 58493, "▁誼": 58494, "▁諌": 58495, "▁諍": 58496, "▁諚": 58497, "▁諛": 58498, "▁諢": 58499, "▁謀": 58500, "▁謁": 58501, "▁謂": 58502, "▁謎": 58503, "▁謡": 58504, "▁謨": 58505, "▁謫": 58506, "▁譏": 58507, "▁譜": 58508, "▁讒": 58509, "▁谺": 58510, "▁豉": 58511, "▁豊": 58512, "▁豕": 58513, "▁豚": 58514, "▁貊": 58515, "▁貟": 58516, "▁販": 58517, "▁貯": 58518, "▁貰": 58519, "▁賈": 58520, "▁賎": 58521, "▁賢": 58522, "▁賭": 58523, "▁贋": 58524, "▁贏": 58525, "▁赦": 58526, "▁赧": 58527, "▁趙": 58528, "▁跏": 58529, "▁跖": 58530, "▁跼": 58531, "▁踆": 58532, "▁踊": 58533, "▁踝": 58534, "▁踰": 58535, "▁蹄": 58536, "▁蹇": 58537, "▁蹲": 58538, "▁蹴": 58539, "▁躇": 58540, "▁躍": 58541, "▁躡": 58542, "▁躱": 58543, "▁軈": 58544, "▁輔": 58545, "▁輙": 58546, "▁輛": 58547, "▁輩": 58548, "▁轡": 58549, "▁轤": 58550, "▁辤": 58551, "▁辷": 58552, "▁迂": 58553, "▁迚": 58554, "▁迦": 58555, "▁逈": 58556, "▁逓": 58557, "▁逝": 58558, "▁逰": 58559, "▁遅": 58560, "▁遖": 58561, "▁遣": 58562, "▁遥": 58563, "▁遮": 58564, "▁遶": 58565, "▁遷": 58566, "▁邈": 58567, "▁郛": 58568, "▁酌": 58569, "▁酖": 58570, "▁酴": 58571, "▁醢": 58572, "▁醪": 58573, "▁醭": 58574, "▁醼": 58575, "▁醿": 58576, "▁釄": 58577, "▁釘": 58578, "▁釧": 58579, "▁釵": 58580, "▁釿": 58581, "▁鈬": 58582, "▁鉄": 58583, "▁鉋": 58584, "▁鉢": 58585, "▁鉦": 58586, "▁銭": 58587, "▁鋒": 58588, "▁鋲": 58589, "▁鋳": 58590, "▁錦": 58591, "▁錫": 58592, "▁鍔": 58593, "▁鍛": 58594, "▁鍮": 58595, "▁鍾": 58596, "▁鎌": 58597, "▁鎗": 58598, "▁鎚": 58599, "▁鏃": 58600, "▁鏐": 58601, "▁鏥": 58602, "▁鐃": 58603, "▁鐫": 58604, "▁鑓": 58605, "▁鑛": 58606, "▁鑵": 58607, "▁鑽": 58608, "▁閏": 58609, "▁閤": 58610, "▁閧": 58611, "▁閨": 58612, "▁閼": 58613, "▁闊": 58614, "▁闖": 58615, "▁阝": 58616, "▁阪": 58617, "▁陵": 58618, "▁隠": 58619, "▁隣": 58620, "▁隺": 58621, "▁雀": 58622, "▁雇": 58623, "▁雌": 58624, "▁雑": 58625, "▁雫": 58626, "▁霜": 58627, "▁霤": 58628, "▁霽": 58629, "▁靖": 58630, "▁靡": 58631, "▁革": 58632, "▁靪": 58633, "▁靭": 58634, "▁鞁": 58635, "▁鞍": 58636, "▁鞠": 58637, "▁韮": 58638, "▁韻": 58639, "▁頊": 58640, "▁頑": 58641, "▁頥": 58642, "▁頴": 58643, "▁頸": 58644, "▁頽": 58645, "▁顋": 58646, "▁顙": 58647, "▁颪": 58648, "▁颭": 58649, "▁颯": 58650, "▁飄": 58651, "▁飢": 58652, "▁飩": 58653, "▁飼": 58654, "▁餓": 58655, "▁餡": 58656, "▁餲": 58657, "▁餹": 58658, "▁餻": 58659, "▁饀": 58660, "▁饂": 58661, "▁饉": 58662, "▁饘": 58663, "▁馨": 58664, "▁駆": 58665, "▁駑": 58666, "▁駕": 58667, "▁駟": 58668, "▁騙": 58669, "▁騫": 58670, "▁驃": 58671, "▁驍": 58672, "▁驥": 58673, "▁驩": 58674, "▁驫": 58675, "▁髄": 58676, "▁髫": 58677, "▁髯": 58678, "▁髱": 58679, "▁髷": 58680, "▁鬌": 58681, "▁鬒": 58682, "▁鬘": 58683, "▁鬛": 58684, "▁鬢": 58685, "▁鬮": 58686, "▁魅": 58687, "▁魑": 58688, "▁魦": 58689, "▁魼": 58690, "▁鮅": 58691, "▁鮇": 58692, "▁鮒": 58693, "▁鮓": 58694, "▁鮗": 58695, "▁鮟": 58696, "▁鮠": 58697, "▁鮥": 58698, "▁鮦": 58699, "▁鮧": 58700, "▁鮬": 58701, "▁鮲": 58702, "▁鮹": 58703, "▁鯊": 58704, "▁鯎": 58705, "▁鯑": 58706, "▁鯒": 58707, "▁鯔": 58708, "▁鯖": 58709, "▁鯯": 58710, "▁鯲": 58711, "▁鯵": 58712, "▁鯷": 58713, "▁鰆": 58714, "▁鰈": 58715, "▁鰊": 58716, "▁鰌": 58717, "▁鰐": 58718, "▁鰒": 58719, "▁鰔": 58720, "▁鰕": 58721, "▁鰛": 58722, "▁鰣": 58723, "▁鰤": 58724, "���鰩": 58725, "▁鰯": 58726, "▁鰶": 58727, "▁鰾": 58728, "▁鱁": 58729, "▁鱅": 58730, "▁鱆": 58731, "▁鱇": 58732, "▁鱊": 58733, "▁鱏": 58734, "▁鱒": 58735, "▁鱚": 58736, "▁鱝": 58737, "▁鱠": 58738, "▁鱣": 58739, "▁鱪": 58740, "▁鱮": 58741, "▁鱲": 58742, "▁鱶": 58743, "▁鱸": 58744, "▁鱺": 58745, "▁鳧": 58746, "▁鳰": 58747, "▁鳽": 58748, "▁鴃": 58749, "▁鴆": 58750, "▁鴨": 58751, "▁鴫": 58752, "▁鴬": 58753, "▁鴻": 58754, "▁鴽": 58755, "▁鵆": 58756, "▁鵉": 58757, "▁鵐": 58758, "▁鵙": 58759, "▁鵺": 58760, "▁鶆": 58761, "▁鶍": 58762, "▁鶚": 58763, "▁鶫": 58764, "▁鷂": 58765, "▁鷃": 58766, "▁鷉": 58767, "▁鷦": 58768, "▁鷯": 58769, "▁鸊": 58770, "▁鸛": 58771, "▁鹵": 58772, "▁麁": 58773, "▁麈": 58774, "▁麌": 58775, "▁麕": 58776, "▁麝": 58777, "▁麹": 58778, "▁黐": 58779, "▁黙": 58780, "▁黽": 58781, "▁黿": 58782, "▁鼇": 58783, "▁鼈": 58784, "▁鼎": 58785, "▁鼫": 58786, "▁鼬": 58787, "▁鼴": 58788, "▁鼺": 58789, "▁齟": 58790, "▁齢": 58791, "▁齬": 58792, "▁龐": 58793, "▁練": 58794, "▁蓮": 58795, "▁隆": 58796, "▁益": 58797, "▁精": 58798, "▁﨟": 58799, "▁僧": 58800, "▁器": 58801, "▁層": 58802, "▁憎": 58803, "▁海": 58804, "▁琢": 58805, "▁祈": 58806, "▁突": 58807, "▁臭": 58808, "▁贈": 58809, "▁難": 58810, "▁－": 58811, "▁Ｏ": 58812, "▁𠀋": 58813, "▁𠵅": 58814, "▁𡈽": 58815, "▁𢦙": 58816, "▁𢳣": 58817, "▁𣇃": 58818, "▁𣉊": 58819, "▁𣏐": 58820, "▁𣴎": 58821, "▁𤇆": 58822, "▁𤋎": 58823, "▁𤍽": 58824, "▁𤸰": 58825, "▁𤹠": 58826, "▁𤾣": 58827, "▁𥙊": 58828, "▁𥝐": 58829, "▁𥡷": 58830, "▁𥶡": 58831, "▁𥸮": 58832, "▁𥿻": 58833, "▁𦒳": 58834, "▁𦚰": 58835, "▁𧝒": 58836, "▁𨉣": 58837, "▁𩞀": 58838, "▁𩷚": 58839, "▁𩹵": 58840, "<s_kuzushiji>": 58841}

config.json

@@ -0,0 +1,24 @@
+{
+  "_name_or_path": "naver-clova-ix/donut-base",
+  "align_long_axis": false,
+  "architectures": [
+    "DonutModel"
+  ],
+  "decoder_layer": 4,
+  "encoder_layer": [
+    2,
+    2,
+    14,
+    2
+  ],
+  "input_size": [
+    960,
+    640
+  ],
+  "max_length": 640,
+  "max_position_embeddings": 640,
+  "model_type": "donut",
+  "torch_dtype": "float32",
+  "transformers_version": "4.11.3",
+  "window_size": 10
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0f95d751953da057de54097688146bd8ef0955bc9d765523ce4d8548860eeff
+size 837068355

sentencepiece.bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb9e3dce4c326195d08fc3dd0f7e2eee1da8595c847bf4c1a9c78b7a82d47e2d
+size 1296245

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}, "additional_special_tokens": ["<s_kuzushiji>"]}

tokenizer_config.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "sp_model_kwargs": {}, "from_slow": true, "name_or_path": "naver-clova-ix/donut-base", "processor_class": "DonutProcessor", "special_tokens_map_file": null, "tokenizer_file": "/home/nsml/.cache/huggingface/transformers/8dff5958cbdb9de4188d643398d5d92bebb82976ce97e6f741b4793e21600485.01bf49938a78cb9ef1792abc3a5829ec39a7887935548bf42fd8d76bf07f15d8", "tokenizer_class": "XLMRobertaTokenizer"}